🦋 Welcome to the MAIN() IRC channel of the Raku Programming Language (raku.org). Log available at irclogs.raku.org/raku/live.html . If you're a beginner, you can also check out the #raku-beginner channel!
Set by lizmat on 6 September 2022.
00:06 yewscion joined 00:10 yewscion left 00:11 yewscion joined 00:20 yewscion left 00:36 simcop2387 left 00:39 yewscion joined 00:47 yewscion left 00:50 yewscion joined 00:56 yewscion left 01:10 yewscion joined 01:17 yewscion left 01:19 yewscion joined 01:24 yewscion left 01:27 kylese left, kylese joined 01:29 yewscion joined 01:33 yewscion left 01:41 yewscion joined 01:46 yewscion left 01:47 yewscion joined 01:52 yewscion left 01:56 kylese left 01:57 kylese joined 02:06 yewscion joined 02:11 yewscion left 02:15 kylese left, kylese joined
SmokeMachine Voldenet: I was playing with your idea and it is very interesting! This is where I got for now: usercontent.irccloud-cdn.com/file/.../image.png 02:26
02:31 yewscion joined 02:39 yewscion left 02:45 yewscion joined 02:53 yewscion left 02:58 Aedil joined
Voldenet SmokeMachine: I like the idea of using field names as actual service names, feels really idiomatic 03:02
without `is keyed<a>` sort of stuff 03:03
03:04 yewscion joined
Voldenet I think it should be possible to do `OAuthClient $oauth is keyed(OAUTH_CLIENT1)` or something similar in case someone wants to use constant variable as service name… 03:06
`OAuthClient $oauth_client1` would prevent you from finding the symbol that easily - you gotta do text search for `oauth_client1` and hope to find all attrs/parameters with the name 03:07
03:08 yewscion left
Voldenet (I've used DI that way and that's the biggest problem with using string as dependencies, bigger than typos or misconfigurations) 03:08
as an aside, containers are not agreeing with how to resolve the missing keyed dependency 03:10
in your example when requesting `Int $b` some containers give you first registered service, some containers give you last, some containers give you first unnamed, some containers give you an error 03:11
it's actually major headache, because it means you can't trivially replace them even if they mostly share the same interface 03:12
what's worse, some libraries/frameworks may require one behavior, different from others 03:13
03:28 yewscion joined 03:40 yewscion left 03:45 kylese left 03:48 kylese joined 03:51 yewscion joined 03:56 yewscion left 04:07 yewscion joined 04:10 Guest19 joined 04:13 arkiuat left 04:15 yewscion left
Guest19 have we considered moving rakudo away from github/microsoft? or the weekly away from wordpress/matt mullenweg? 04:24
04:27 arkiuat joined 05:04 arkiuat left 05:33 arkiuat joined 05:37 yewscion joined 05:38 arkiuat left 05:43 yewscion left 05:48 yewscion joined 05:54 yewscion left 05:57 yewscion joined 06:02 yewscion left 06:07 arkiuat joined 06:12 arkiuat left 06:13 yewscion joined 06:22 yewscion left 06:27 Sgeo left 06:42 yewscion joined 06:43 arkiuat joined 06:47 arkiuat left 06:51 yewscion left 06:54 arkiuat joined 07:23 apac joined 07:37 Guest15 joined, apac left 07:55 dakkar joined 07:57 Guest15 left 07:59 arkiuat left 08:01 Guest19 left 08:06 yewscion joined 08:11 yewscion left 08:23 lichtkind joined 08:26 yewscion joined 08:30 arkiuat joined 08:31 yewscion left 08:35 arkiuat left 08:39 Aedil left 08:42 yewscion joined 08:51 yewscion left 08:53 arkiuat joined 08:58 arkiuat left 09:07 yewscion joined 09:12 yewscion left
wayland76 Guest19: To what? 09:20
09:23 yewscion joined 09:27 arkiuat joined
melezhik. Sourcehut, codeberg , Gitlab - not too many alternatives. But GH still has advantages not presented there, imho . 09:30
09:31 arkiuat left 09:33 yewscion left, yewscion joined
librasteve there is definitely a desire to move from Wordpress since Matt imploded (for me) … I build Wordpress sites for money and HARC is born out of “this sucks, I like to code (in raku), there must be a better way” … initially HARC is going to check off static sites (think updated raku.org) so blog is a bit downstream since I want to write my text in markdown and i want to have raku code highlighting plus will need 09:39
authentication which needs forms (which is current wip) and CRUD
09:39 melezhik joined
SmokeMachine Voldenet: I think `has $.oauth is named-on-deps<blablabla>` would make sense and should be possible (or anything similar). How should be defined what to do with a not exact match? Something like this? `has $.bla is dying-on-unnamed-dep`, `has $.ble is first-unnamed-dep`, `has $.bli is last-unnamed-dep`. Does something like that make sense? Or should it be defined on the container? 09:39
melezhik Testing my irc client … 09:40
librasteve otoh, I treat GH as a necessary evil for now … but one day an alternate would be nicer
09:42 yewscion left 09:45 arkiuat joined 09:58 yewscion joined 10:03 yewscion left 10:09 yewscion joined 10:14 yewscion left 10:29 wayland76 joined, wayland left 10:52 arkiuat left 11:11 arkiuat joined 11:16 arkiuat left
Voldenet SmokeMachine: I don't know, there hasn't been a single good aproach to this I've seen – I've seen it done as switchable container defaults (it obviously breaks library expectations) or per-dependency config (pointless, because not idiomatic) 11:30
oh, also I've seen per-parameter configurators (if you see class X having parameter Y, use this dependency, otherwise use that), it helps resolving some problems which are headache to figure out 11:34
I guess the latter is fine, but nobody uses named dependencies soon enough 11:35
11:38 arkiuat joined 11:47 yewscion joined 12:03 yewscion left, nine left 12:04 nine joined 12:26 yewscion joined 12:31 yewscion left 12:35 yewscion joined
antononcube GitHub is way too convenient. Also, it renders Jupyter notebooks. From 2019 to 2022 I have been "forced" to use GitLab and Bitbucket, I they were not providing that many features as GitHub. (They might now, but I am not that curious to try them.) 12:38
12:41 yewscion left
SmokeMachine Voldenet: I forgot to show how my try is working about mro and roles... for now it looks like this: usercontent.irccloud-cdn.com/file/.../image.png 13:29
Voldenet: do you mind if I continue playing with that? I'm having a lot of fun. And do you mind if I keep asking your opinion? 13:31
librasteve yeah this was the sadya nutella way … kill with kindness 13:42
Xliff \o 13:43
What happens during zef STAGING phase?
SmokeMachine ???
Xliff What operations are performed?
SmokeMachine I may have missed something... 13:45
SmokeMachine is confused
Xliff zef performs a staging phase during install. I was wondering what happens in that phase. 13:46
This may be a ugexe question. 13:47
SmokeMachine Voldenet: maybe I'm going too far... but I thought about this: usercontent.irccloud-cdn.com/file/.../image.png 14:05
14:06 yewscion joined
SmokeMachine any opinion about that ☝️? 14:06
14:11 yewscion left
Voldenet SmokeMachine: I don't mind when people have fun in general :> 14:28
SmokeMachine Voldenet: I mean, I don't want to still your ideas... :)
steal? rob... 14:29
14:30 yewscion joined
SmokeMachine (sorry... may English is too bad...) 14:30
Voldenet you can't steal them, because they're already stolen from various DI implementers ;p
but yeah, I admit that on top of being useful, this allows a lot of room for experiments 14:31
SmokeMachine in portugese we say that a robber that steals from another robber has 100 years of pardon ("ladrao que rouba ladrao tem 100 anos de perdao") 14:32
Voldenet :)
SmokeMachine Voldenet: do you have an opinion about my extrapolation? (I mean my last image) 14:33
Voldenet it actually makes sense to have scoping simplified like that, but 14:34
there should be a way to use two scopes at the same time 14:35
it's actually very useful, because it allows you to move or copy data between scopes
14:35 yewscion left
SmokeMachine could you point me to an example of that? 14:36
Voldenet: usercontent.irccloud-cdn.com/file/.../image.png 14:41
14:42 yewscion joined 14:47 yewscion left 14:52 Aedil joined
SmokeMachine Voldenet: import-deps adds a container as parent of the current conteiner: usercontent.irccloud-cdn.com/file/.../image.png 14:56
15:02 arkiuat_ joined 15:04 arkiuat left 15:06 yewscion joined
Voldenet SmokeMachine: example what I have in mind (of course "new-scope" is not implemented) 0x0.st/8xIs.raku 15:25
I didn't even syntax-check it, but I hope it's more or less clear what should be possible 15:29
also, one thing is pretty important - scopes should be implicit in most cases 15:30
so, background thread should have its scope, request also should etc. 15:31
I think assigning the scope to deps could work if you skip the fact that scopes dispose - this is actually important for things like db connections 15:32
so `deps { … }` should mean `$di.new-scope; …; $di.dispose` - it's fine in most cases 15:33
the "explicit" management should be used only in cases like the example - where you really want to dance with multiple containers and do stuff 15:34
SmokeMachine but isn't enough to left the scope to it be destroyed? 15:36
Voldenet nope, GC with finalization queue can cause a lot of db/http connections being alive 15:37
and if there's not much memory pressure, they can survive for a long time
so, when request ends, scope should have something like '.cleanup', that would automatically run all the cleanups registered in scope 15:38
since we're in DI world, it's easy to define `register(Http::Client, -> Cleanups $x { my $c = Http::Client.new; $x.on-cleanup({ $c.close-all }); $c })` 15:40
it's even worse when you do connection pooling, 100 connections is plenty, but if you don't release them back to the pool, you'll going to exhaust the pool quicker than DESTROY can kick in 15:43
s/you'll/you're/
but with di automatically doing cleanup after every request: `register(Connection, -> ConnectionPool $p, Cleanups $x { my $c = $p.rent; $x.on-cleanup({ $p.release }); $c });` would solve the problem 15:45
erm, it should be $c.release 15:46
maybe more idiomatic than .new-scope/.cleanup would be .enter-scope, .on-leave, .leave 15:47
and in fact: `register(Connection, -> ConnectionPool $p, Scope $s { my $c = $p.rent; $s.on-leave({ $p.release }); $c });` - using "Scope" as di variable could effectively mark its lifetime as Scope-bound 15:48
SmokeMachine or maybe if using `deps {...}`, maybe it could use `deps { LEAVE do-release; ... }` ? 15:57
16:37 dakkar left
SmokeMachine Voldenet: maybe I went too far again: usercontent.irccloud-cdn.com/file/.../image.png 16:47
17:01 RakuGuest joined 17:08 apac joined 17:30 RakuGuest left 18:02 librasteve_ joined
jubilatious1_98524 Matt? 18:13
librasteve www.reddit.com/r/Wordpress/comment...down_from/ 18:17
SmokeMachine Voldenet: github.com/FCO/Deps 18:24
librasteve SmokeMachine: thanks for the lesson in traits ... hopefully my latest commit works for you 18:28
SmokeMachine :) I'm running the tests now 18:39
librasteve: merged! 18:40
18:45 MoC joined
antononcube Cannot make much sense of it. Good to know, anyway. 18:48
librasteve SmokeMachine: thanks! 19:03
@antononcube there is no sense in what Matt is doing with Wordpress - time to bail 19:04
antononcube My WordPress blogs are too "popular", so, I will be very slow at that... 🙂 19:05
librasteve well seems like we will need to export / import ... rome wasn't built in a day - my advice is to have your own domain name 19:40
20:16 Aedil left 20:33 yewscion left 21:01 MoC left
SmokeMachine Is there a way to changes a sub’s signature wrapping it? 21:03
21:06 yewscion joined
SmokeMachine m: sub a($a) { say $a }; &a.wrap: sub () { nextwith 42 }; a # I mean something like this 21:07
camelia ===SORRY!=== Error while compiling <tmp>
Calling a() will never work with declared signature ($a)
at <tmp>:1
------> $a }; &a.wrap: sub () { nextwith 42 }; <HERE>a # I mean something like this
SmokeMachine m: sub a($a?) { say $a }; &a.wrap: sub () { nextwith 42 }; a # if I make it optional, it works… but I didn’t want that… 21:08
camelia 42
21:12 yewscion left 21:15 apac left 21:37 yewscion joined 21:39 Xliff left 21:42 yewscion left 22:06 yewscion joined 22:12 yewscion left 22:43 yewscion joined 22:47 yewscion left 22:55 lichtkind left 23:04 melezhik left 23:05 wayland76 left 23:15 yewscion joined 23:20 yewscion left 23:29 yewscion joined 23:34 yewscion left
ab5tract It should at least work for this case: 23:36
m: sub a($a) { say $a }; &a.wrap: sub (--> Int) { nextwith 42 }; a()
camelia ===SORRY!=== Error while compiling <tmp>
Calling a() will never work with declared signature ($a)
at <tmp>:1
------> a.wrap: sub (--> Int) { nextwith 42 }; <HERE>a()
ab5tract that is, when the sub signature guarantees a value 23:37
SmokeMachine: would you mind making a ticket for this?
23:39 Sgeo joined 23:52 yewscion joined
SmokeMachine ab5tract: github.com/rakudo/rakudo/issues/5890 23:56

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!